Doubao Seed-2.1-pro tested: 1M context handles 3D sites, 61-sheet Excel and real research
卡尔的AI沃茨 · wechat · 2026-09-16
Hands-on review of Doubao-Seed-2.1-pro-0915 (1M context, 256k deep thinking) across five real tasks: rebuilding 3D interactive sites with Three.js, processing a 61-sheet engineering Excel (correctly handling a missing sheet), and a sourced Toronto office-location report with a 31-item appendix. Also introduces the 'QBS method' of turning books into skills to curb AI hallucinated answers.
More from coding & agent
- Building a classifier in 2023 meant hacking OpenAI's function calling API — how far we've come — _ScottCondron · 2026-09-16
- Agent memory isn't one thing: sessions, wiki, procedural, and the forgotten 'judgment' layer — sujingshen · 2026-09-16
- Dev: wiring all service providers into MCP lets agents debug issues — DanielLockyer · 2026-09-16
- An agent with full-key access is the real risk: four rules for least-privilege agent credentials — sujingshen · 2026-09-16
- Agents slow down as memory notes pile up, unlike humans who speed up with practice — sujingshen · 2026-09-16
- 5 Open-Source AI Agent Repos Exploding on GitHub Right Now — Shruti_0810 · 2026-09-16